Therapies currently in preclinical development for prion disease seek to lower prion protein (PrP) expression in the brain. Trials of such therapies are likely to rely on quantification of PrP in cerebrospinal fluid (CSF) as a pharmacodynamic biomarker and possibly as a trial endpoint. Studies using PrP ELISA kits have reproducibly shown that CSF PrP is lowered in the symptomatic phase of disease, a potential confounder for reading out the effect of PrP-lowering drugs in symptomatic patients. To date it has been unclear whether the reduced abundance of PrP in CSF results from its incorporation into plaques, retention in intracellular compartm ents, downregulation as a function of the disease process, or other factors. Because misfolding or proteolytic cleavage could potentially render PrP invisible to ELISA even if its concentration were constant or increasing in disease, we sought to establish an orthogonal method for CSF PrP quantification. We developed a targeted mass spectrometry method based on multiple reaction monitoring (MRM) of nine PrP tryptic peptides quantified relative to known concentrations of isotopically labeled standards. Analytical validation experiments showed process replicate coefficients of variation below 15%, good dilution linearity and recovery, and suitable performance for both CSF and brain homogenate and across humans as well as preclinical species of interest. In N=55 CSF samples from individuals referred to prion surveillance centers with rapidly progressive dementia, all six human PrP peptides, spanning the N-and C-terminal domains of PrP, were uniformly reduced in prion disease cases compared to individuals with non-prion diagnoses. This confirms the findings from ELISA studies, demonstrating that lowered CSF PrP concentration in prion disease is a genuine result of the disease process and not merely an artifact of ELISA-based measurement. We provide a targeted mass spectrometry-based method suitable for preclinical and clinical quantification of CSF PrP as a tool for drug development.